Region duplication blind detection based on multiple feature combination

The paper presents an efficient approach based on feature combination for detecting the region splice image by copy-paste manipulation. The combined features include 1D moment, 2D moment and Markov feature, which could efficiently capture the most representative features. Each block used for test is represented by the combined feature, and the spliced region is detected by feature match. Experiments on Columbia Image Splicing Detection Evaluation Dataset demonstrates that the copy-paste forgery regions could be accurately detected by the presented method.